CIEN301 – Numerical Analysis

Numerical Differentiation and Integration

1
• Differentiation and integration are core parts of calculus.
• Differentiation – Definition:
A difference approximation of differentiation is:
∆𝑦 𝑓 𝑥𝑖 + ∆𝑥 − 𝑓 𝑥𝑖
=
∆𝑥 ∆𝑥
• If ∆𝑥 is allowed to approach zero, the difference becomes a derivative:
𝑑𝑦 𝑓 𝑥𝑖 + ∆𝑥 − 𝑓 𝑥𝑖
= lim
𝑑𝑥 ∆𝑥→0 ∆𝑥
• The second derivative – commonly known as the curvature - represents the
derivative of the first derivative, which indicates how fast the slope is changing.
𝑑2 𝑦 𝑑 𝑑𝑦
2
=
𝑑𝑥 𝑑𝑥 𝑑𝑥
2
The graphical definition of a derivative: as ∆𝑥 approaches
zero, the difference becomes a derivative

3
The Partial Derivative
• The partial derivative is equivalent to taking the derivative of the
function at a point with all but one variable held constant. For a
function 𝑓(𝑥, 𝑦) dependent on both 𝑥 and 𝑦, the partial derivative of
f with respect to 𝑥 at an arbitrary point (𝑥, 𝑦) is defined as:
𝜕𝑓 𝑓 𝑥 + ∆𝑥, 𝑦 − 𝑓 𝑥, 𝑦
= lim
𝜕𝑥 ∆𝑥→0 ∆𝑥
• Similarly, the partial derivative of 𝑓 with respect to 𝑦 is defined as:
𝜕𝑓 𝑓 𝑥, 𝑦 + ∆𝑦 − 𝑓 𝑥, 𝑦
= lim
𝜕𝑦 ∆𝑦→0 ∆𝑦
• The partial derivative corresponds to slopes of a point on a surface in
the 𝑥 or 𝑦 directions.

4
 Integration
• Integration is the inverse process to
differentiation.
• Mathematically, integration is
represented by:
𝑏

𝐼 = න 𝑓 𝑥 𝑑𝑥
𝑎
where, 𝑓(𝑥) is referred to as the
integrand and 𝑎 & 𝑏 are the limits of
integration.
• The integral 𝐼 represents the area under
the curve of 𝑓(𝑥) between 𝑥 = 𝑎 and 𝑏.

5
 Differentiation vs. Integration
• Given that 𝑦 𝑡 and 𝑣 𝑡
represent position and
velocity of a given object at
time 𝑡, respectively, then,
• 𝑣 𝑡 is the derivative of 𝑦 𝑡
with respect to 𝑡.
𝑑
𝑣 𝑡 = 𝑦 𝑡
𝑑𝑡
• Conversly, the position of the
object 𝑦(𝑡) can be
determined as integration of
𝑣(𝑡).
𝑡

𝑦 𝑡 = න 𝑣 𝑡 𝑑t
0

6
Mathematical Background

7
Mathematical Background

8
Integration
• Two main approaches of numerical integration will be discussed:
• Integration methods based on replacing a complicated function or
tabulated data with a simple polynomial – Newton-Cotes formulas.
These includes: trapezoidal rule, Simpson’s 1/3 rule and Simpson’s
3/8 rule.
• Integration methods designed to integrate known equations and
functions:
• Romberg Integration
• Adaptive Integration
• Gauss Quadrature

9
Newton-Cotes Integration Formulas
• The Newton-Cotes formulas are the most common numerical integration
schemes.
• They are based on the strategy of replacing a complicated function or
tabulated data with an approximating function that is easy to integrate.
• The approach involves approximating the integration of a function 𝑓(𝑥)
with a polynomial 𝑓𝑛 𝑥 , i.e.
𝑏 𝑏

𝐼 = න 𝑓 𝑥 𝑑𝑥 ≅ න 𝑓𝑛 𝑥 𝑑𝑥
𝑎 𝑎
Where the polynomial of order 𝑛, 𝑓𝑛 𝑥 , is expressed as:
𝑓𝑛 𝑥 = 𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 + ⋯ + 𝑎𝑛−1 𝑥 𝑛−1 + 𝑎𝑛 𝑥 𝑛
10
Approximations of a function integration with a that of a
polynomial

• (a) First-order
(trapezoidal)
• (b) Single parabola
(Simpons’s 1/3 rule)

11
Approximating an integral with a series of polynomials

• Integrals can be
approximated using a series
of polynomials applied
piecewise to the function or
data over equal-length
segments.

12
 Closed vs. Open forms of the Newton-Cotes formulas

(a) Closed forms are those where

the data points at the
beginning and end of the
limits of integration are known
(focus of this course)

(b) Integration limits extend

beyond the range of the data
(similar to extrapolation)

13
 The Trapezoidal Rule
• The trapezoidal rule is the first of the
Newton-Cotes integration formulas.
• The data or the function is approximated
with a first-order polynomial, i.e.
𝑏 𝑏

𝐼 = න 𝑓 𝑥 𝑑𝑥 ≅ න 𝑓1 𝑥 𝑑𝑥
𝑎 𝑎
• This integration can be approximated by
the following formula – known as the
trapezoidal rule:
𝑓 𝑎 +𝑓 𝑏
𝐼 = 𝑏−𝑎
2

14
The Truncation Error of the Trapezoidal Rule
• The truncation error of the trapezoidal rule is:
1 ′′
𝐸𝑡 = − 𝑓 𝜉 𝑏 − 𝑎 3
12
where 𝜉 is an intermediate value between a and b.
• The truncation error of the trapezoidal rule is of the order of ℎ3 ,
where
ℎ =𝑏−𝑎

15
 Example of a Single Application of the Trapezoidal Rule
• Use the trapezoidal rule to numerically integrate
the following function from 𝑎 = 0 to 𝑏 = 0.8
(the exact value of the integral can be calculated
analytically and is found to be 1.640533)
𝑓 𝑥 = 0.2 + 25𝑥 − 200𝑥 2 + 675𝑥 3 − 900𝑥 4 + 400𝑥 5
𝑓 𝑎 +𝑓 𝑏
The trapezoidal rule: 𝐼 = 𝑏 − 𝑎
2
𝑓 𝑎 = 0 = 0.2; 𝑓 𝑏 = 0.8 = 0.232
Substitute into the trapezoidal rule:
0.2 + 0.232
𝐼 ≅ 0.8 − 0 = 0.1728
2
The actual error can be calculated as:
𝐸𝑡 = 1.640533 − 0.1728 = 1.467733
𝜖 𝑡 = 89.5%
16
 The Multiple-Application Trapezoidal Rule
• The accuracy of the trapezoidal rule can be improved by dividing
the integration interval from a to b into equal number of
segments and apply the method to each segment.
• The integral would then be equal to the sum of the individual
segments. The resulting equations are called multiple-
application, or composite, integration formulas.
• A general equation for the trapezoidal rule can be expressed as
𝑓 𝑥0 + 2 σ𝑛−1
𝑖=1 𝑓 𝑥𝑖 + 𝑓 𝑥𝑛
𝐼=ℎ
2
𝑏−𝑎
• Where ℎ = is the segment size
𝑛
• And the truncation error
𝑛
𝑏−𝑎 3 𝑏 − 𝑎 3 𝑏 − 𝑎 3
𝐸𝑡 = − ෍ 𝑓 ′′ 𝜉 ≅ − 𝑛 𝑓 ҧ′′ = − 𝑓 ҧ′′
12𝑛3 12𝑛3 12𝑛2
𝑖=1
Thus, if the number of segments is doubled, the truncation error
will be quartered.

17
 Example of a Multiple-Application Trapezoidal Rule
• Use the two-segment trapezoidal rule to numerically integrate the following function from
𝑎 = 0 to 𝑏 = 0.8 (Recall from the previous example that the exact value of the integral
1.640533)
𝑓 𝑥 = 0.2 + 25𝑥 − 200𝑥 2 + 675𝑥 3 − 900𝑥 4 + 400𝑥 5
The general trapezoidal rule:
𝑓 𝑥0 + 2 σ𝑛−1
𝑖=1 𝑓 𝑥𝑖 + 𝑓 𝑥𝑛
𝐼=ℎ
2
Substitute into the trapezoidal rule for 𝑛 = 2:
0.8 − 0 𝑓 0 + 2 × 𝑓 0.4 + 𝑓 0.8 0.2 + 2 × 2.456 + 0.232
𝐼= × = 0.4 = 1.0688
2 2 2
The actual error can be calculated as:
𝐸𝑡 = 1.640533 − 1.0688 = 0.57173
𝜖 𝑡 = 34.9%

18
Truncation error diminishes as ℎ gets smaller
The following results are obtained from applying
multiple-application trapezoidal rule to the
previous example with decreasing segment size.

19
 Simpson’s Rules
• A more accurate estimate of the integral can be obtained by using higher-order
polynomials to connect the points.
• For example, if there is an extra point midway between f(a) and f(b), the three
points can be connected with a parabola (Simpson’s 1/3 rule).
• If there are two points equally spaced between f(a) and f(b), the four points can
connected with a third-order polynomial (Simpson’s 3/8 rule).

20
Segments (n)
Comparing Newton-Cotes Formulas (single application)

Newton-Cotes Formula Simplified Newton-Cotes

Name h
Formula
Points

Trapezoidal 𝑓 𝑥0 + 𝑓 𝑥1 𝑏−𝑎 𝑓 𝑥0 + 𝑓 𝑥1
1 2 Rule ℎ
1
𝑏−𝑎
2 2
Simpson’s 1/3 1 𝑏−𝑎 𝑓 𝑥0 + 4𝑓 𝑥1 + 𝑓 𝑥2
2 3 rule 3
ℎ 𝑓 𝑥0 + 4𝑓 𝑥1 + 𝑓 𝑥2
2
𝑏−𝑎
6
Simpson’s 3/8 3 𝑏−𝑎 𝑓 𝑥0 + 3𝑓 𝑥1 + 3𝑓 𝑥2 + 𝑓 𝑥3
3 4 rule 8
ℎ 𝑓 𝑥0 + 3𝑓 𝑥1 + 3𝑓 𝑥2 + 𝑓 𝑥3
3
𝑏−𝑎
8

• Writing the Newton-Cotes formulas in terms of ℎ shows where the names of

the rules are derived from.
21
 Example of a Single Application of Simpson’s 1/3 Rule
• Use the Simpson’s 1/3 rule to numerically integrate the same function from the previous
example (𝑎 = 0 to 𝑏 = 0.8) (as before the exact answer is 1.640533)
𝑓 𝑥 = 0.2 + 25𝑥 − 200𝑥 2 + 675𝑥 3 − 900𝑥 4 + 400𝑥 5
𝑓 𝑥0 +4𝑓 𝑥1 +𝑓 𝑥2
The Simpson’s 1/3 rule: 𝐼 = 𝑏 − 𝑎
6
𝑎+𝑏 0+0.8
𝑥0 = 𝑎 = 0; 𝑥1 = = = 0.4; 𝑥2 = 𝑏 = 0.8
2 2

Substitute into the Simpson’s 1/3 rule:

𝑓 0 + 4𝑓 0.4 + 𝑓 0.8 0.2 + 4 × 2.456 + 0.232
𝐼 ≅ 0.8 = 0.8 = 1.367467
6 6
The actual error can be calculated as:
𝐸𝑡 = 1.640533 − 1.367467 = 0.2730667 → 𝜖 𝑡 = 16.6%
Compare with the truncation for the trapezoidal rule (𝜖 𝑡 = 89.5%). The Simpson’s 1/3 is 5
times more accurate than the trapezoidal rule.
22
 The Multiple-Application Simpson’s 1/3 Rule
• The Simpson’s rule can be improved by
dividing the integration interval into a
number of segments of equal width
𝑏−𝑎
ℎ=
𝑛
• The total integral can be represented as
𝑥2 𝑥4 𝑥𝑛

𝐼 = න 𝑓 𝑥 𝑑𝑥 + න 𝑓 𝑥 𝑑𝑥 + ⋯ + න 𝑓 𝑥 𝑑𝑥
𝑥0 𝑥2 𝑥𝑛−2

23
 The Multiple-Application Simpson’s 1/3 Rule
• Substituting Simpson’s 1/3 rule for the individual
integral yields:
𝑓 𝑥0 +4𝑓 𝑥1 +𝑓 𝑥2 𝑓 𝑥2 +4𝑓 𝑥3 +𝑓 𝑥4
• I ≅ 2ℎ + 2ℎ +⋯+
6 6
𝑓 𝑥𝑛−2 +4𝑓 𝑥𝑛−1 +𝑓 𝑥𝑛
2ℎ
6

• Combing the terms yield the following general equation:

𝑓 𝑥0 + 4 σ𝑛−1
𝑖=1,3,5 𝑓 𝑥𝑖 + 2 σ 𝑛−2
𝑗=2,4,6 𝑓 𝑥𝑗 + 𝑓 𝑥𝑛
𝐼=ℎ
3

• Notice multiple application of Simpson’s 1/3 rule can be

employed only if the number of segments is even.

24
 Example of a Multiple-Application Simpson’s 1/3 Rule
• Use the multiple-application Simpson’s 1/3 rule with 𝑛 = 4 to numerically integrate the
same function from previous examples (𝑎 = 0 to 𝑏 = 0.8) (recall from the previous
example that the exact value of the integral 1.640533)
𝑓 𝑥 = 0.2 + 25𝑥 − 200𝑥 2 + 675𝑥 3 − 900𝑥 4 + 400𝑥 5
𝑏−𝑎 0.8−0
For 𝑛 = 4: ℎ = = = 0.2
4 4

𝑥0 = 𝑎 = 0; 𝑥1 = 𝑥0 + ℎ = 0.2; 𝑥2 = 0.4; 𝑥3 = 0.6 ; 𝑥4 = 0.8

Substitute into the multiple-application Simpson’s 1/3 rule:
𝑓 𝑥0 + 4 𝑓 𝑥1 + 𝑓 𝑥3 + 2𝑓 𝑥2 + 𝑓 𝑥4
𝐼=ℎ
3
0.2 + 4 1.288 + 3.464 + 2 2.456 + 0.232
𝐼 = 0.2 = 1.623467
3
𝐸𝑡 = 1.640533 − 1.623467 = 0.017067
𝜖 𝑡 = 1.04%

25
 Comparing Truncation Errors of Newton-Cotes Formulas
(single application)
Segments (n)
Points

Newton-Cotes Formula
Name h Truncation Errors

𝑏−𝑎 3
𝑓 𝑥0 + 𝑓 𝑥1 𝑏−𝑎
1 2 Trapezoidal Rule ℎ
1
𝐸𝑡 = − 𝑓 ′′ 𝜉
2 12
1 𝑏−𝑎 𝑏−𝑎 5 4
2 3 Simpson’s 1/3 rule
3
ℎ 𝑓 𝑥0 + 4𝑓 𝑥1 + 𝑓 𝑥2
2
𝐸𝑡 = − 𝑓 𝜉
2880
3 𝑏−𝑎 𝑏−𝑎 5 4
3 4 Simpson’s 3/8 rule
8
ℎ 𝑓 𝑥0 + 3𝑓 𝑥1 + 3𝑓 𝑥2 + 𝑓 𝑥3
3
𝐸𝑡 = − 𝑓 𝜉
6480

• The trapezoidal rule has the largest truncation error.

• The truncation errors of Simpson’s 1/3 rule and Simpson’s 3/8 rule are of the same
order.
26
 Deciding which integration formula to use.
• Simpson’s 1/3 rule is the method of preference since it
attains a level of accuracy only slightly less than that of
Simpson’s 3/8 rule using only three points rather than
4 points.
• However, the 3/8 version could be required in
situations where the number of segments is odd.
• Consider for example that an integration estimate is
required for five segments. In this case, multiple
application of Simpson’s 1/3 can not be done.
• If you choose a multiple-application of the trapezoidal
rule, you will get a large truncation error.
• A better approach is to apply the Simpson’s 1/3 rule to
the first two segments and Simpson’s 3/8 rule to the
last three.
27
 Example of a Single Application of Simpson’s 3/8 Rule
• Use the Simpson’s 3/8 rule to numerically integrate the same function from previous examples (𝑎 = 0 to 𝑏 =
0.8) (Recall from the previous example that the exact value of the integral 1.640533)
𝑓 𝑥 = 0.2 + 25𝑥 − 200𝑥 2 + 675𝑥 3 − 900𝑥 4 + 400𝑥 5
For a single application of Simpson’s 3/8 rule, the number of segments 𝑛 = 3, consequently
𝑏−𝑎 0.8 − 0
ℎ= = = 0.2667
3 3
𝑥0 = 0; 𝑥1 = 𝑥0 + ℎ = 0.2667; 𝑥2 = 𝑥0 + 2ℎ = 0.5333; 𝑥3 = 0.8

Substitute into the single-application Simpson’s 3/8 rule:

3
𝐼 = ℎ 𝑓 0 + 3𝑓 0.2667 + 3𝑓 0.5333 + 𝑓 0.8 =
8
3
𝐼 = ℎ 0.2 + 3 1.432724 + 3 3.487177 + 0.232
8
𝐸𝑡 = 1.640533 − 1.519170 = 0.1213630

𝜖 𝑡 = 7.4%

28
 Example of using Simpson’s 3/8 Rule with Simpson’s 1/3 Rule
• Use the Simpson’s 3/8 rule in conjunction with Simpson’s 1/3 rule to numerically integrate
the same function from previous examples (𝑎 = 0 to 𝑏 = 0.8) for five segments (recall from
the previous example that the exact value of the integral 1.640533)
𝑓 𝑥 = 0.2 + 25𝑥 − 200𝑥 2 + 675𝑥 3 − 900𝑥 4 + 400𝑥 5
𝑏−𝑎 0.8
The data needed for a five-segment application (ℎ = = = 0.16
𝑁 5

𝑥0 = 0; 𝑥1 = 0.16; 𝑥2 = 0.32; 𝑥3 = 0.48; 𝑥4 = 0.64; 𝑥5 = 0.80

The integral for the first two segments is obtained using Simpson’s 1/3 rule:
1
𝐼 = ℎ 𝑓 𝑥0 + 4𝑓 𝑥1 + 𝑓 𝑥2
3
1 1
𝐼 = 0.16 𝑓 0 + 4𝑓 0.16 + 𝑓 0.32 = 0.16 0.2 + 4 1.296919 + 1.743393
3 3
= 0.3803237

29
 Example of using Simpson’s 3/8 Rule with Simpson’s 1/3 Rule
For the last three segments (𝑥2 𝑡𝑜 𝑥3 ;𝑥3 𝑡𝑜 𝑥4 ;𝑥4 𝑡𝑜 𝑥5 ), the 3/8 rule can be used to
obtain
3
ℎ 𝑓 𝑥2 + 3𝑓 𝑥3 + 3𝑓 𝑥4 + 𝑓 𝑥5 =
8
3
0.16 𝑓 0.32 + 3𝑓 0.48 + 3𝑓 0.64 + 𝑓 0.8 =
8
3
0.16 1.743393 + 3 3.186015 + 3 3.181929 + 0.232 = 1.264754
8
The total integral is computed by summing the two results:
𝐼 = 0.3803237 + 1.264753 = 1.645077
𝐸𝑡 = 1.640533 − 1.645077 = −0.00454383
𝜖 𝑡 = −0.28%
The truncation error is much lower than those of earlier applications.
30
Integration with Unequal Segments
• The previous formulas discussed earlier apply only for segments of
equal size, i.e. equally spaced data points.
• One approach of numerically integrating unevenly spaced data is to
apply the trapezoidal data to each segment and sum the results:
𝑓 𝑥0 + 𝑓 𝑥1 𝑓 𝑥1 + 𝑓 𝑥2 𝑓 𝑥𝑛−1 + 𝑓 𝑥𝑛
𝐼 = ℎ1 + ℎ2 + ⋯ + ℎ𝑛
2 2 2
where ℎ𝑖 =the width of segment 𝑖.
• If some adjacent segments are equal then they could be evaluated
using Simpson’s rules.

31
 Example: Trapezoidal Rule with Unequal Segments
• Consider the following table generated with unequally
x f(x)
spaced values based on the polynomial used in the previous 0.00 0.200000
examples. 0.12 1.309729
𝑓 𝑥 = 0.2 + 25𝑥 − 200𝑥 2 + 675𝑥 3 − 900𝑥 4 + 400𝑥 5 0.22 1.305241
0.32 1.743393
• Determine based on the trapezoidal rule the integral for this
data. (recall that correct answer is 1.640533) 0.36 2.074903
0.40 2.456000
• Applying the trapezoidal rule yields: 0.44 2.842985
𝐼 = 0.12
0.2 + 1.309729
+ 0.10
1.309729 + 1.305241
+ ⋯ + 0.1
2.363 + 2.363 0.54 3.507297
2 2 2
= 0.090584 + 0.130749 + ⋯ + 0.12975 = 1.594801 0.64 3.181929
0.70 2.363000
which represents an absolute percent relative error of 𝜀𝑡 = 2.8%.
0.80 0.232000
32
Example: Inclusion of Simpson’s Rules in the Evaluation of
Uneven Data
• Recompute the integral for the same data, but use x f(x)
Simpson’s rules for those segments where they are 0.00 0.200000
appropriate.
0.12 1.309729
• The first segment is evaluated with the trapezoidal 0.22 1.305241
rule:
0.2 + 1.309729 0.32 1.743393
𝐼 = 0.12 − 0.0 = 0.09058376
2 0.36 2.074903
• The next two segments (from x=0.12 to 0.32) are of 0.40 2.456000
equal length. Consequently their integral can be 0.44 2.842985
computed with a single application of Simpson’s 1/3
rule: 0.54 3.507297
1.309729 + 4 1.305241 + 1.743393 0.64 3.181929
𝐼 = 0.32 − 0.12
6 0.70 2.363000
= 0.2726863
0.80 0.232000
33
 Example: Inclusion of Simpson’s Rules in the Evaluation of
Uneven Data – contd.
• The next three segments (from x=0.32 to 0.44) are of
x f(x)
equal length. Consequently their integral can be 0.00 0.200000
computed with a single application of Simpson’s 3/8 0.12 1.309729
rule: 0.22 1.305241
𝑓 0.32 + 3𝑓 0.36 + 3𝑓 0.40 + 𝑓 0.44 0.32 1.743393
𝐼 = 0.44 − 0.32 0.36 2.074903
8
1.743393 + 3 2.074903 + 3 2.456 + 2.842985 0.40 2.456000
𝐼 = 0.12
8 0.44 2.842985
= 0.2726863
0.54 3.507297
0.64 3.181929
0.70 2.363000
0.80 0.232000
34
 Example: Inclusion of Simpson’s Rules in the Evaluation of
Uneven Data – contd.
• The next two segments (from 𝑥 = 0.44 𝑡𝑜 0.64) are
x f(x)
of equal length. Consequently their integral can be 0.00 0.200000
computed with single application of Simpson’s 1/3 0.12 1.309729
rule: 0.22 1.305241
𝑓 0.44 + 4𝑓 0.54 + 𝑓 0.64 0.32 1.743393
𝐼 = 0.64 − 0.44 0.36 2.074903
6
2.842985 + 4 3.507297 + 3.181929 0.40 2.456000
𝐼 = 0.64 − 0.44 0.44 2.842985
6
= 0.6684701 0.54 3.507297
0.64 3.181929
0.70 2.363000
0.80 0.232000
35
 Example: Inclusion of Simpson’s Rules in the Evaluation of
Uneven Data – contd.
• Finally, the last two segments, which are of unequal
x f(x)
length, can be evaluated with the trapezoidal rule to 0.00 0.200000
give values of 0.1663479 and 0.1297500, respectively. 0.12 1.309729
0.22 1.305241
• The area of these individual segments can be
summed to yield a total integral: 0.32 1.743393
0.36 2.074903
0.09058376 + 0.2758029 + 0.2726863
0.40 2.456000
+ 0.6684701 + 0.1663479 + 0.1297500
= 1.603641 0.44 2.842985
0.54 3.507297
Which represents an error of 𝜀𝑡 = 2.2%, which is 0.64 3.181929
superior to the result using the trapezoidal rule in the 0.70 2.363000
previous example.
0.80 0.232000
36
Example: Inclusion of Simpson’s Rules in the Evaluation of
Uneven Data – contd.

37
 Multiple Integrals
• Multiple integrals – widely common in
engineering - involve more than one variable.
• For example, a general equation to compute
the average of a two-dimensional function can
be written as:
𝑑 𝑏
‫𝑥 𝑓 𝑎׬ 𝑐׬‬, 𝑦 𝑑𝑥 𝑑𝑦
𝑓ҧ =
𝑑−𝑐 𝑏−𝑎
The numerator is called a double integral.
• From calculus (order of integration is not
important
𝑑 𝑏 𝑏 𝑑
න න 𝑓 𝑥, 𝑦 𝑑𝑥 𝑑𝑦 = න න 𝑓 𝑥, 𝑦 𝑑𝑦 𝑑𝑥
𝑐 𝑎 𝑎 𝑐

39
 Example: Using double integral to determine average
temperature
• The temperature of a rectangular
heated plate is described by the
following function:
𝑇 𝑥, 𝑦 = 2𝑥𝑦 + 2𝑥 − 𝑥 2 − 2𝑦 2 + 72
• Compute the average temperature
given that the plate is 8-m long (𝑥
dimension) and 6-m wide (𝑦
dimension).
• Note that the simple average is 47.33
• The true value can be obtained
analytically and is equal to 58.66667

40
 Example: Using double integral to determine average
temperature
Solution:
• Use two-segment applications of the trapezoidal rule in each dimension.
𝑓 𝑥0 +2 σ𝑛−1
𝑖=1 𝑓 𝑥𝑖 +𝑓 𝑥𝑛
𝐼= ℎ (general multiple trapezoidal))
2
𝑏−𝑎 𝑓 𝑥0 +2𝑓 𝑥1 +𝑓 𝑥2
For two-segment trapezoidal rule 𝐼 =
2 2

• Apply the trapezoidal rule along the 𝑥 dimension for each y values. Then
these values are integrated along the 𝑦 dimension to give a final result equal
to 2688.
• Dividing this number by the area yields the average temperature as
2688Τ 6 × 8 = 56.

41
42
Example: Using double integral to determine average
temperature
• A single-segment Simpson’s 1/3 rule can be applied in the same
fashion.
• This results in an integral of 2816 and average of 58.66667, which is
exact.
• This occurs because Simpson’s 1/3 rule is based on 2nd order
polynomial and will therefore produce a perfect match for 2nd order
(cubic) polynomials.
• Since the highest order term in the function is 2nd order, the same
exact result occurs for present case.

43